feat: Python File interface types support

Author: thewtex

src/python/itkwasm/itkwasm/binary_file.py

@@ -1,6 +1,6 @@
 from dataclasses import dataclass
+from pathlib import PurePosixPath
 
 @dataclass
 class BinaryFile:
-    data: bytes
-    path: str
+    path: PurePosixPath

src/python/itkwasm/itkwasm/pipeline.py

@@ -1,11 +1,13 @@
 import json
 from pathlib import Path
-from typing import List, Union, Dict
+from typing import List, Union, Dict, Tuple
 from .interface_types import InterfaceTypes
 from .pipeline_input import PipelineInput
 from .pipeline_output import PipelineOutput
 from .text_stream import TextStream
 from .binary_stream import BinaryStream
+from .text_file import TextFile
+from .binary_file import BinaryFile
 
 from wasmer import engine, wasi, Store, Module, ImportObject, Instance
 from wasmer_compiler_cranelift import Compiler
@@ -26,9 +28,21 @@ def __init__(self, pipeline: Union[str, Path, bytes]):
         self.module = Module(self.store, wasm_bytes)
         self.wasi_version = wasi.get_version(self.module, strict=True)
 
-    def run(self, args: List[str], outputs: List[PipelineOutput]=[], inputs: List[PipelineInput]=[], preopen_directories=[], map_directories={}, environments={}) -> List[PipelineOutput]:
+    def run(self, args: List[str], outputs: List[PipelineOutput]=[], inputs: List[PipelineInput]=[]) -> Tuple[PipelineOutput]:
         """Run the itk-wasm pipeline."""
 
+        preopen_directories=set()
+        map_directories={}
+        # Todo: expose?
+        environments={}
+        for index, input_ in enumerate(inputs):
+            if input_.type == InterfaceTypes.TextFile or input_.type == InterfaceTypes.BinaryFile:
+                preopen_directories.add(str(input_.data.path.parent))
+        for index, output in enumerate(outputs):
+            if output.type == InterfaceTypes.TextFile or output.type == InterfaceTypes.BinaryFile:
+                preopen_directories.add(str(output.data.path.parent))
+        preopen_directories = list(preopen_directories)
+
         wasi_state = wasi.StateBuilder('itk-wasm-pipeline')
         wasi_state.arguments(args)
         wasi_state.environments(environments)
@@ -60,6 +74,10 @@ def run(self, args: List[str], outputs: List[PipelineOutput]=[], inputs: List[Pi
                 array_ptr = self._set_input_array(data_array, index, 0)
                 data_json = { "size": len(data_array), "data": f"data:application/vnd.itk.address,0:{array_ptr}" }
                 self._set_input_json(data_json, index)
+            elif input_.type == InterfaceTypes.TextFile:
+                pass
+            elif input_.type == InterfaceTypes.BinaryFile:
+                pass
             else:
                 raise ValueError(f'Unexpected/not yet supported input.type {input_.type}')
 
@@ -80,13 +98,17 @@ def run(self, args: List[str], outputs: List[PipelineOutput]=[], inputs: List[Pi
                     data_size = self.output_array_size(0, index, 0)
                     data_array = bytes(memoryview(self.memory.buffer)[data_ptr:data_ptr+data_size])
                     output_data = PipelineOutput(InterfaceTypes.BinaryStream, BinaryStream(data_array))
+                elif output.type == InterfaceTypes.TextFile:
+                    output_data = PipelineOutput(InterfaceTypes.TextFile, TextFile(output.data.path))
+                elif output.type == InterfaceTypes.BinaryFile:
+                    output_data = PipelineOutput(InterfaceTypes.BinaryFile, BinaryFile(output.data.path))
                 populated_outputs.append(output_data)
 
         delayed_exit = instance.exports.itk_wasm_delayed_exit
         delayed_exit(return_code)
 
         # Should we be returning the return_code?
-        return populated_outputs
+        return tuple(populated_outputs)
 
     def _set_input_array(self, data_array: bytes, input_index: int, sub_index: int) -> int:
         data_ptr = 0

src/python/itkwasm/itkwasm/text_file.py

@@ -1,6 +1,6 @@
 from dataclasses import dataclass
+from pathlib import PurePosixPath
 
 @dataclass
 class TextFile:
-    data: str
-    path: str
+    path: PurePosixPath

src/python/itkwasm/test/input/input.bin

@@ -0,0 +1 @@
+ޭ��

src/python/itkwasm/test/input/input.txt

@@ -0,0 +1 @@
+The answer is 42.

src/python/itkwasm/test/test_pipeline.py

@@ -1,8 +1,9 @@
-from pathlib import Path
+from pathlib import Path, PurePosixPath
 
-from itkwasm import InterfaceTypes, TextStream, BinaryStream, PipelineInput, PipelineOutput, Pipeline
+from itkwasm import InterfaceTypes, TextStream, BinaryStream, PipelineInput, PipelineOutput, Pipeline, TextFile, BinaryFile
 
 test_input_dir = Path(__file__).resolve().parent / 'input'
+import tempfile
 
 
 def test_stdout_stderr():
@@ -16,7 +17,7 @@ def test_pipeline_bytes():
     pipeline = Pipeline(wasm_bytes)
     pipeline.run([])
 
-def test_pipeline_input_output_files():
+def test_pipeline_input_output_streams():
     pipeline = Pipeline(test_input_dir / 'input-output-files-test.wasi.wasm')
 
     pipeline_inputs = [
@@ -45,4 +46,45 @@ def test_pipeline_input_output_files():
     assert outputs[1].data.data[0], 222
     assert outputs[1].data.data[1], 173
     assert outputs[1].data.data[2], 190
-    assert outputs[1].data.data[3], 239
+    assert outputs[1].data.data[3], 239
+
+def test_pipeline_input_output_files():
+    pipeline = Pipeline(test_input_dir / 'input-output-files-test.wasi.wasm')
+    input_text_file = PurePosixPath(test_input_dir / 'input.txt')
+    input_binary_file = PurePosixPath(test_input_dir / 'input.bin')
+
+    with tempfile.TemporaryDirectory() as tmpdirname:
+        output_text_file = PurePosixPath(Path(tmpdirname) / 'output.txt')
+        output_binary_file = PurePosixPath(Path(tmpdirname) / 'output.bin')
+
+        pipeline_inputs = [
+            PipelineInput(InterfaceTypes.TextFile, TextFile(input_text_file)),
+            PipelineInput(InterfaceTypes.BinaryFile, BinaryFile(input_binary_file)),
+        ]
+
+        pipeline_outputs = [
+            PipelineOutput(InterfaceTypes.TextFile, TextFile(output_text_file)),
+            PipelineOutput(InterfaceTypes.BinaryFile, BinaryFile(output_binary_file)),
+        ]
+
+        args = [
+            '--use-files',
+            '--input-text-file', str(input_text_file),
+            '--input-binary-file', str(input_binary_file),
+            '--output-text-file', str(output_text_file),
+            '--output-binary-file', str(output_binary_file)
+        ]
+
+        outputs = pipeline.run(args, pipeline_outputs, pipeline_inputs)
+
+        assert outputs[0].type == InterfaceTypes.TextFile
+        with open(outputs[0].data.path, 'r') as fp:
+            content = fp.read()
+            assert content == 'The answer is 42.'
+        assert outputs[1].type, InterfaceTypes.BinaryFile
+        with open(outputs[1].data.path, 'rb') as fp:
+            content = fp.read()
+            assert content[0] == 222
+            assert content[1] == 173
+            assert content[2] == 190
+            assert content[3] == 239